Publication Details

Deepfake Speech Detection: A Spectrogram Analysis

FIRC Anton, MALINKA Kamil and HANÁČEK Petr. Deepfake Speech Detection: A Spectrogram Analysis. In: SAC '24: Proceedings of the 39th ACM/SIGAPP Symposium on Applied Computing. 2024, p. 9.
Czech title
Detekce deepfake řeči: Analýza spektrogramů
Type
conference paper
Language
english
Authors
Firc Anton, Ing. (DITS FIT BUT)
Malinka Kamil, Mgr., Ph.D. (DITS FIT BUT)
Hanáček Petr, doc. Dr. Ing. (DITS FIT BUT)
Keywords

Deepfake, Speech, Image-based, Deepfake Detection, Spectrogram

Abstract

The current voice biometric systems have no natural mechanics to defend against deepfake spoofing attacks. Thus, supporting these systems with a deepfake detection solution is necessary. One of the latest approaches to deepfake speech detection is representing speech as a spectrogram and using it as an input for a deep neural network. This work thus analyzes the feasibility of different spectrograms for deepfake speech detection. We compare types of them regarding their performance, hardware requirements, and speed. We show the majority of the spectrograms are feasible for deepfake detection. However, there is no general, correct answer to selecting the best spectrogram. As we demonstrate, different spectrograms are suitable for different needs.

Published
2024 (in print)
Pages
9
Proceedings
SAC '24: Proceedings of the 39th ACM/SIGAPP Symposium on Applied Computing
Conference
ACM Symposium On Applied Computing, Avila, ES
BibTeX 
@INPROCEEDINGS{FITPUB12908,
   author = "Anton Firc and Kamil Malinka and Petr Han\'{a}\v{c}ek",
   title = "Deepfake Speech Detection: A Spectrogram Analysis",
   pages = 9,
   booktitle = "SAC '24: Proceedings of the 39th ACM/SIGAPP Symposium on Applied Computing",
   year = 2024,
   language = "english",
   url = "https://www.fit.vut.cz/research/publication/12908"
}
Back to top